Best Gold Scalper EA for MT4/MT5

You searched for the best gold scalper EA because you want your XAUUSD trading automated. That part is reasonable. Gold moves every day, it respects technical levels that a machine can read, and no human wants to sit in front of a chart at 3am for a 40-point move.

The problem is not that good gold EAs do not exist. The problem is that you cannot tell them apart. Every list you open ranks something different, every page ends with a buy button, and almost nothing on the page tells you what the robot does when a trade goes wrong — which is the only thing that actually matters.

You are one search away from fifty gold scalper EAs that all look the same

You have already seen the pages. Same layout. Same simulated equity curve climbing from bottom-left to top-right. Same “99% modelling quality” screenshot. Same testimonials from first names. Different logo.

So you do the sensible thing and look for reviews. Now you are deeper in. One page calls a gold EA a scam, another calls the same EA a goldmine, and both of them are affiliates for a different robot. Somewhere in the middle there is a Telegram channel showing balance screenshots with no deposit history attached.

Then you start reading the sales pages properly, and the language gives the game away. “Consistent monthly returns.” “Survives every market condition.” “Fully hands-free.” Not one of them tells you the maximum drawdown in the worst month of the last two years, and not one of them shows the lot size sequence when the first two trades lose.

Here is the uncomfortable part: this is not a research failure on your part. The information you need is usually not published. The performance screenshots come from the vendor. The backtest was run with the vendor’s own settings on the vendor’s own data. The compiled file you are asked to run has no readable code inside it. You are being asked to trust a stranger with your capital and offered nothing to inspect.

And gold punishes that specific mistake more than most markets do. XAUUSD is not a clean, cheap instrument to trade mechanically. Its spread is one of the widest of any major symbol after the exotic crosses, and it is not stable. On a quiet London morning you might see 15 or 20 points round trip. Ten minutes into a US inflation release, that same spread can be 80, 120, or wider, briefly, while price is moving faster than anything else on your watchlist. A scalper that averages 40 points per winning trade cannot absorb that. It is not a matter of skill or tuning. The arithmetic does not work.

So there are three ways a gold scalper EA takes your money:

1. It doubles down. A losing trade is not closed, it is averaged with a bigger position. This works often enough to look clever, until one trending day in gold goes further than the model assumed.
2. It is fitted to the past. The parameters were chosen because they produced a beautiful curve on one specific slice of history. Change the broker, the spread or the year and the edge disappears.
3. It is too cheap to run. A $6-a-month VPS with 400ms of latency, on a broker 3,000 miles away, turns a planned entry into an accidental one.

None of those three show up in a screenshot.

What a bad gold scalper cost him

Here is an illustrative composite, built from a pattern this kind of product repeats. Daniel is not a real customer.

Daniel saves $2,400 for a gold EA. Not a fortune. Enough that losing it hurts. The vendor charges $299, and before he pays, he does the demo run: two weeks on a demo account, 11% up, 74 trades, one small losing stretch. He feels good. He has done the test everyone recommends.

He funds a live account with $2,000 and runs the default settings on a VPS he bought for $11 a month. Week one: +$61. Week two: -$34. Week three, a Wednesday with a Federal Reserve statement, gold moves hard for 90 minutes. His equity goes from $2,020 to $1,480 in the space of an hour and a half. Not because the EA had no stop — because it had no stop on the basket. Each leg had been opened as a fresh position with its own take-profit, and the losing leg was reopened at a bigger size each time.

He closes everything by hand at 4:52pm. He is down $540. He is also sitting on a screenshot that the vendor’s Telegram channel posted that evening with the caption “another strong day, our clients are smiling.”

Here is what that episode actually cost:

• $299 for the licence, gone.
• $540 of trading losses, which understates the risk badly — a 13% drawdown on a corrective day could easily have been 40% on a trend day.
• Four months of attention, spent arguing with the vendor’s support group chat, reading about chargebacks and re-reading the sales page to see what he missed.

The money is the small part. The expensive part is what he now believes: that automated trading is a con, and that the two years he spent learning price structure were wasted.

They were not. The mistake was method, not market. He judged a gold EA by its headline return and its demo run — the two numbers that a bad EA is built to produce. Nobody gave him a list of checks that a bad EA fails.

That is what the rest of this page is.

What actually works when you automate gold scalping

The mechanism you need is boring. You want to run a gold scalper whose entry logic is visible, whose worst case is capped by a stop-loss on every single position, and whose risk per trade is set by you rather than by the developer. Then you want to test it yourself on real ticks and on demo before a single live dollar is exposed.

You can buy that. You can also get it for free, and because this page carries a download, that is where we start — with a real, GPL-licensed open-source robot whose full MQL source you can read line by line.

What is the EA31337 multi-strategy robot?

It is an open-source Expert Advisor for MetaTrader 4 and MetaTrader 5, written in MQL and published on GitHub by the EA31337 project under the GPL-3.0 licence. Rather than hard-coding one entry rule, it bundles a large set of individual strategies — moving averages, oscillators, breakout and other classics — that you switch on or off, and each can read its own timeframe. It will trade XAUUSD like any other symbol your broker lists, but it is a general robot, not a gold-only scalper, and that is an honest limit to keep in mind.

The risk layer is the reason it is on this page. The strategies sit on top of a shared money-management and order layer, so the same sizing and stop rules apply to every strategy you enable. Position size comes from a configurable risk setting rather than a fixed lot, and the order layer is where the stop is placed. Account-level guards are inputs you set, and the project documents them instead of hiding them inside a binary.

None of those features are exotic. What matters is that you can verify them. The project ships as MQL source. Open it in any editor and search for how the lot size is calculated, how the stop is placed, and what happens after a loss. If you do not like any of it, change it and recompile. That is not possible with a compiled binary.

The project publishes its own backtest and optimisation notes, and because it is popular it has been run on many brokers by other traders. Treat every published number as the authors’ own claims, not as evidence for your account. They have not been audited on your feed, your spread or your commission, and a result on one broker does not transfer to another. The reason to download this file is not the headline return. It is that you get a complete, readable, risk-capped open-source robot to test against your own broker and your own spread.

Two other honest limits. It is a large framework with a real learning curve — you configure which strategies run and on what, so it is not a single preset you attach and forget. And it is a general-purpose tool: nothing about it is tuned for gold’s spread and volatility, so treat it as a testbed for the six checks below rather than a specialist gold scalper.

Why does a readable source file beat a compiled gold EA?

Because a compiled Expert Advisor is a black box that you are handing your account to. A compiled file contains no readable logic. You cannot see the lot sequence, you cannot see whether a losing position is ever left without a stop, and you cannot see whether the entries look ahead at future bars, which is the most common way a backtest gets flattered.

What you are checking	Open-source source code (this download)	Compiled file from a vendor
Lot size calculation	Readable, verifiable, editable	Hidden
Stop-loss on every position	Visible in the order request	Claims only
Martingale or grid logic	Search for it in seconds	Undetectable
Look-ahead in entry logic	Auditable bar by bar	Undetectable
Cost	Free (GPL-3.0 licence)	Usually a paid licence; prices vary
Support	Community issues, no SLA	Usually included
Who fixes a bug	You can	The vendor, if they still trade the account

None of that makes an open-source EA better at predicting gold. It makes it checkable. For a system that will hold positions while you sleep, checkable is worth more than polished.

Which six numbers separate a gold scalper from a lottery ticket?

Ask for these six numbers, and ask for them from a third-party tracker rather than from the vendor’s own page. If a vendor cannot or will not produce them for a real-money account, that is your answer.

Metric	What to look for	Why it decides the outcome
Maximum drawdown	Under 20% of account equity, measured peak to trough	It is the number that determines whether you can stay in the trade long enough for the edge to appear
Profit factor	Above 1.2 over at least 200 trades	Below 1.2, spread and slippage variations can erase the whole edge
Total trade count	At least 200, ideally 500+	A scalper with 30 recorded trades has no statistical story, only an anecdote
Average trade duration	Under 60 minutes for a scalper	If it holds for days, it is not scalping, and your risk model is wrong
Worst losing streak	Compare it to your account size and leverage	A 12-trade losing run at 2% risk per trade is a 24% drawdown in sequence
Cost per trade	Spread plus commission in points	This is the EA’s rent. On gold it is often the difference between positive and negative

Read the drawdown first, and read it in currency, not as a percentage of the vendor’s demo balance. A 15% drawdown on a $200,000 demo is a very different experience to 15% on the $2,000 you actually funded.

Why is martingale the one feature that should disqualify a gold EA?

Because it converts a known, bounded loss into an unknown, unbounded one.

Here is the mechanism. A martingale gold scalper takes a trade and loses. Instead of closing that loss and moving on, it opens another position in the same direction at a larger size, hoping a small retrace recovers both. If price keeps going, it opens a third, larger still. The equity curve stays flat and comfortable for weeks, because each recovery pushes the whole ladder back to breakeven plus a little. Then gold trends, the ladder grows past what the margin can carry, and the account is closed out at the worst possible price.

There is a second layer of danger specific to gold. Retail leverage on gold is capped at 20:1 in the EU under the European Securities and Markets Authority’s product intervention measures — 30:1 on major currency pairs and 20:1 for gold and major indices. That cap exists precisely because leveraged positions in volatile instruments destroy retail accounts. A martingale system is a machine for building a position that the leverage cap was designed to prevent.

How to spot it without reading code:

• An input named LotMultiplier, Multiplier, RecoveryFactor or GridStep.
• A maximum-positions setting of 10 or more.
• A statement with an unusual pattern: many small wins, no single loss larger than the wins.
• Any sentence on the sales page like “recovers losing trades” or “no losing months”.

The shape is the tell. A real scalper’s statement has clusters of losses. A martingale statement looks like a savings account until it doesn’t.

How much does spread and slippage cost a gold scalper per month?

More than most people calculate before they buy. Run the numbers on your own broker, because they change the verdict.

Take a scalper that places 10 trades a day, 20 days a month, at a 25-point round-trip spread on XAUUSD with 0.01 lots. That is 200 trades and 5,000 points of spread paid in a month. On XAUUSD quoted to two decimals, one point is one cent per ounce, and a 0.01 lot is one ounce. So each 25-point round trip costs about $0.25. Scale the same 200 trades to 0.10 lots and the spread bill is about $500, a quarter of a $2,000 account. Lot size sets the cost as much as the spread does.

Trades per month	Round-trip spread	Spread cost on 0.01 lots	Spread cost on 0.10 lots
50	25 points	~$12.50	~$125
100	25 points	~$25	~$250
200	25 points	~$50	~$500
200	15 points	~$30	~$300

Now the same logic on a demo account, where the spread is often tighter than your live account will be. This is why a demo that looks flat at best can look strong. The MetaTrader 5 documentation is explicit that during testing the spread is not modelled but taken from historical data, so the backtest inherits whatever spread values the history file carries — which is usually your broker’s older, better spread. Before you trust any gold backtest, check what the spread assumptions were.

Practical rules:

• Set a maximum spread filter in the EA and make it realistic. If your broker quotes 20 points at 10am London and 90 at 2:30pm New York, a 30-point filter kills most of your New York trades — which may be the right outcome.
• Trade the sessions where the cost-to-range ratio is best. London and the New York open carry the volume; the late Asian session on gold is usually thin.
• Count commission as part of the spread, not separately. Every broker eventually charges it in some form.

MT4 or MT5 for gold scalping?

If you are starting fresh with gold automation, use MetaTrader 5. The reason is testing, not trading. MetaTrader 5’s Strategy Tester supports real tick data from your broker and the “Every tick” generation mode, and the MQL5 documentation is blunt about why that matters: the faster “1 minute OHLC” mode can produce what it calls a “Testing Grail” — a beautiful curve that a strategy cannot reproduce in live conditions. If your gold scalper’s results look unrealistically smooth on a rough testing mode, the documentation’s own advice is to re-run it in “Every tick” mode.

MetaTrader 4 is still fine for live execution and has the larger library of legacy gold EAs. But it gives you less ability to prove that a strategy worked for the right reason. If you are choosing between the two, our comparison of the best MT4 EAs covers where MT4 still makes sense, and the geraked MT5 expert advisors collection is a useful second open-source reference to test alongside this one.

Can a gold scalper EA pass a prop firm challenge in 2026?

It can, but the constraints are tighter than most EA marketing suggests, and the pass depends more on the risk inputs than on the strategy.

Here are the actual FTMO numbers, from their published trading objectives. On the 2-Step programme the targets are 10% then 5%, with a 5% maximum daily loss and a 10% maximum loss, and a minimum of four trading days. On the 1-Step programme the target is 10%, the maximum daily loss is 3%, the maximum loss is 10% on a trailing end-of-day basis, and there is a Best Day Rule that caps your best day at 50% of total positive days’ profit.

That arithmetic is unforgiving for a scalper running default settings:

Risk per trade	Three losses in one day	Against FTMO 1-Step daily limit (3%)
0.5%	1.5%	Survives
1.0%	3.0%	Reaches the limit
2.0%	6.0%	Breaches badly

This is why the number that matters on a funded account is not the profit factor. It is the daily loss limit, and the discipline to stop trading when it is hit. Our gold prop firm robot guide goes deeper on presets and programme rules, and the top ranking tables compare funded programmes directly.

If you are testing a gold EA on a funded account, run the same EA on a free trial first with the exact risk inputs you intend to use. The trial is free and the lesson is inexpensive.

Exactly what you get in this download

The download is the complete repository: the MQL source tree, the licence, and the project’s documentation. You are not getting a compiled binary, because you do not want one.

Component	Detail
File	The MQL source tree (compile it yourself in MetaEditor)
Platform	MetaTrader 4 and MetaTrader 5
Instrument	Any symbol your broker lists, including XAUUSD
Strategy set	Many independent strategies you enable or disable, each able to read its own timeframe
Risk inputs	A configurable risk percentage, plus account-level guards
Filters	Per-strategy and order-layer conditions that decide whether a signal becomes a trade
Trade management	Shared sizing, stop placement and order handling across the enabled strategies
Dependencies	Standard MetaTrader includes — no external DLLs
Licence	GPL-3.0. Free to use and modify; anything you distribute must stay under the same licence
Author	EA31337 project
Cost	Free

What it does not do

This is the section that most download pages skip, and it is the one worth reading.

• It does not include a compiled binary. You compile it in MetaEditor. That takes thirty seconds and means you know exactly which code is running.
• It is not a gold-only scalper. It is a general multi-strategy robot. It will trade XAUUSD, but nothing in it is tuned to gold’s spread and volatility, so expect to configure and test it rather than attach and forget it.
• It does not configure itself. You choose which strategies run, on which symbols and timeframes. The default set is a starting point, not a finished strategy.
• It does not have a support desk. It is a community project with a public repository. Bug reports go to the project or to your own editor.
• It does not come with a verified live track record for your broker. Any published results are the authors’. Your spread, commission and server time are different, so no third party has confirmed them on your account.
• It does not guarantee anything about your result. No EA can. The market you are trading it on is the same market that has bankrupted professional desks.
• It does not detect your broker’s symbol. You have to check that XAUUSD is spelled the same way in Market Watch as the chart you attach it to. Brokers use XAUUSD.m, GOLD or XAUUSD_raw.
• It does not fix a bad account size. Below a few hundred dollars, the minimum lot and the spread cost dominate everything the strategy does.

If you want a licensed, supported alternative with presets and a developer behind it, the site’s own gold scalpers are worth reading about on their product pages: Zenith Matrix EA, Nexora Manus EA, Mythos Epic EA and Onix Stratos XAUUSD EA. The trade-off is honest and symmetrical: you get support and tested presets, and you give up the ability to read the code. Read the specifications, check the refund terms, and apply the same six checks above to them as to anything else.

Questions traders ask before they put a gold EA on a chart

Do I need a VPS? For live gold scalping, yes, at least while you are evaluating whether the EA works. Gold scalpers are sensitive to latency because the targets are small. A cheap VPS in a data centre near your broker’s trading server costs a few dollars a month and removes a variable you cannot otherwise control. On demo, latency matters less, so test the logic first and move to a VPS when you go live.

Which session should a gold scalper trade? The London session and the first two hours of New York carry the volume and the cleanest directional moves in gold. That is why you should restrict the robot to those two windows yourself and leave the Asian session off. If you enable the Asian session, expect fewer valid signals, because the ATR and spread filters you set will reject most of them on a thin book.

What lot size should I start with? Whatever the risk setting produces, and never larger. Set the robot’s risk percentage to 1.0 and a 1% loss is the target if its stop is hit. On a $1,000 account that is $10, which is near the minimum lot on some brokers anyway — check the broker’s minimum lot and its value per point before you assume the setting is doing what you expect.

Can I run two gold EAs on the same account? Technically yes, and it is a common way to double your daily drawdown without noticing. Two EAs each risking 1% per trade can put 4% at risk on the same hour, and if they trade the same signal you have effectively doubled your position, not diversified. If you run more than one, set a combined daily loss cap and check that the magic numbers differ.

What should I do when the EA hits a losing streak? Nothing, until the streak exceeds the worst losing run in your test data. That threshold is the reason you should log it before you start. Interfering with a system that is behaving within its tested range is how a planned 8% drawdown becomes a 25% one, because the position you closed manually was the trade that was going to recover the account.

Is MetaTrader 5’s Strategy Tester enough on its own? No. It is a filter, not a verdict. It tells you whether the logic is coherent and whether the risk controls work. It cannot tell you how the EA handles your broker’s real spreads at 2:30pm on a payrolls day, and that is exactly the scenario that shows up in the drawdown.

What to do next, in the next ten minutes

Download the repository. Compile the source. Put it on a XAUUSD demo chart with the risk percentage at 1.0 and a spread cap set to something your broker actually quotes outside news. Run one year of real ticks in the Strategy Tester. Then leave it on demo for a fortnight and watch how it behaves on the day gold moves 3% while you are not paying attention.

That fortnight tells you more about gold scalping than any review page can, including this one. You will see the spread widen and the EA stand aside. You will see a losing run and be forced to decide, in advance, what your response is. You will find out what your worst day looks like in currency rather than in percent.

Then, and only then, decide whether this robot or any other deserves real money.

For more free, licence-verified tools to run alongside it, the free download library is the place to start, and EA31337 Libre is a widely used open-source multi-strategy EA if you want a second codebase to study. If you would rather compare licensed options first, the product pages carry the specifications and the pricing.

Trading forex and CFDs on gold carries a high risk of loss and is not suitable for everyone. Leverage magnifies both directions. Past performance, including every backtest and every track record on this page, does not indicate future results. Test on a demo account first, and only ever risk money you can afford to lose.